Existence of weak adiabatic limit in almost all models of perturbative QFT Paweł Duch Jagiellonian University, Cracow, Poland LQP 40 Foundations and Constructive Aspects of Quantum Field Theory, 23.06.2017 1 / 16
The Wightman and Green functions are one of the most important objects in the quantum field theory in the Minkowski space. The perturbative definition of the Green functions in a large class of models was given by Lowenstein (1976) and Breitenloher Maison (1977). Using the Epstein-Glaser approach both the Wightman and Green functions can be defined. ù One has to show the existence of the weak adiabatic limit. The existence of the weak adiabatic limit has been proved so far in purely massive models, the quantum electrodynamics and the massless ϕ 4 theory. Main result The existence of the weak adiabatic limit in a large class of models in the Minkowski space including all models with the interaction vertices of the canonical dimension equal 4. ñ The perturbative construction of the Wightman and Green functions. ñ The definition of a Poincaré invariant functional on the algebra of the interacting fields. 2 / 16
Plan of the talk 1. Axioms of the time-ordered products. 2. Definition of the Wightman and Green functions in the Epstein-Glaser approach. 3. Known and new results about the existence of the adiabatic limit. 4. Outline of the proof. 3 / 16
Epstein-Glaser approach the notation Scalar, spinor and vector fields. Massive or massless fields. Only renormalizable models. The notation: The basic generators: A 1,..., A p. The generators: B α A i, α - a multi-index The algebra of the symbolic fields =the free unital commutative algebra generated by B α A i. Monomials: A r ś i,α pbα A iq rpi,αq. The super-quadri-index: r : t1,..., pu ˆ N 4 Q pi, αq ÞÑ rpi, αq P N. Polynomials: B ř r arar, a r P C. Sub-polynomials: B psq ř r Wick polynomials: :Bpxq:. r! pr sq! arar s, s - super-quadri-index. 4 / 16
Epstein-Glaser approach the time-ordered products TpB 1,..., B nqpx 1,..., x nq TpB 1px 1q,..., B npx nqq : SpR 4n q Ñ LpD 0q 1. TpHq 1, TpBpxqq :Bpxq: and TpB 1px 1q,..., B npx nq, 1px n`1qq TpB 1px 1q,..., B npx nqq. 2. Symmetry: TpB 1px 1q,..., B npx nqq TpB πp1q px πp1q q,..., B πpnq px πpnq qq. 3. Translational covariance: Upaq TpB 1px 1q,..., B npx nqqupaq 1 TpB 1px 1 ` aq,..., B npx n ` aqq. 4. Causality: For x 1,..., x m Á x m`1,... x n it holds TpB 1px 1q,..., B npx nqq TpB 1px 1q,..., B mpx mqq TpB m`1px m`1q,..., B npx nqq. 5. Wick expansion: TpB 1px 1q,..., B npx nqq ÿ pω TpB ps1q 1 px 1q,..., Bn psnq px nqqωq :As 1 px 1q... A sn px nq:. s 1!... s n! s 1,...,s n 6. Bound on the Steinmann s scaling degree: n`1 ÿ sdp pω TpB 1px 1q,..., B npx nq, B n`1p0qqωq q ď pdimpb jq ` cq j 1 5 / 16
Epstein-Glaser approach interacting models The interaction vertices L 1,..., L q, the coupling constants e 1,..., e q and the switching functions g 1,..., g q P SpR 4 q, g : pg 1,..., g qq. Interacting fields with IR regularization Bogliubov s formula ż qÿ ż Spg; hq Texp i d 4 x e jg jpxql jpxq ` i d 4 x hpxqbpxqpxq j 1 δ B retpg; xq : p iq δhpxq Spg; 0q 1 Spg; hqˇ (1) ˇh 0 (2) Time-ordered products of interacting fields with IR regularization ż qÿ ż mÿ Spg; hq Texp i d 4 x e jg jpxql jpxq ` i d 4 x h jpxqb jpxqpxq j 1 j 1 TpB 1,retpg; x 1q,..., B m,retpg; x mqq: p iq m δ δh... δ mpx mq δh Spg; 0q 1 Spg; hqˇ 1px 1q Wightman and Green functions with IR regularization (3) ˇh 0 (4) pω B 1,retpg; x 1q... B m,retpg; x mqωq (5) pω TpB 1,retpg; x 1q,..., B m,retpg; x mqqωq (6) 6 / 16
The adiabatic limit 1. The Wightman and Green functions are well-defined as formal power series in e 1,..., e q as long as all the switching functions belong to the Schwartz class. To make physical predictions one has to take the limit g 1pxq,..., g qpxq Ñ 1. 2. For any g P SpR N q such that gp0q 1 we define a one-parameter family of Schwartz tests functions: g ɛpxq : gpɛxq for ɛ ą 0. (7) We have lim ɛœ0 g ɛpxq 1 pointwise, lim ɛœ0 g ɛpqq p2πq N δpqq in S 1 pr N q. 3. Let t P S 1 pr N q and consider the limit ż ż lim d N x tpxqg ɛpxq lim ɛœ0 ɛœ0 d N q p2πq N tpqq g ɛp qq c. (8) If the above limit exists and its value is independent of the choice of g P SpR N q such that gp0q 1 then we say that the adiabatic limit of t exists and equals c and the distribution t has the value c at zero in the sense of Łojasiewicz. 7 / 16
Known results about the existence of the adiabatic limit Epstein Glaser (1973) the weak adiabatic limit The existence of the weak adiabatic limit in purely massive theories: WpB 1px 1q,..., B mpx mqq : lim ɛœ0 pω B 1,retpg ɛ; x 1q,..., B m,retpg ɛ; x mqωq (9) GpB 1px 1q,..., B mpx mqq : lim ɛœ0 pω TpB 1,retpg ɛ; x 1q,..., B m,retpg ɛ; x mqqωq (10) The above limits are taken in S 1 pr 4m q. Blanchard Seneor (1975) the weak adiabatic limit The existence of the weak adiabatic limit in the quantum electrodynamics and the massless ϕ 4 theory. Epstein Glaser (1976) the strong adiabatic limit The existence of the S-matrix in purely massive theories: SΨ : lim ɛœ0 Spg ɛqψ for all Ψ P D 1. (11) Fredenhagen Lindner (2014) The existence of expectation values of the products of the interacting fields in thermal states. 8 / 16
Existence of the weak adiabatic limit Theorem Assume that the interaction vertices L 1,..., L q of a given model satisfy one of the following conditions: 1. c 0, dimpl l q 4 for all l, 2. c 1, dimpl l q 3 and L l contains at least one massive field for all l. ñ It is possible to normalize the time-ordered products such that the weak adiabatic limit exists (the explicit form of the required normalization condition is stated on the next slide). The bound on the Steinmann s scaling degree implies that sdp pω T `L ps 1q l 1 px 1q,..., L psnq l n px nq, L ps n`1q l n`1 p0q Ωq q ď ω 4n, (12) where ω : 4 řp i 1rdimpAiq epaiq ` dpaiqs is a function of s1,..., sn`1, epa iq # the external lines corresponding to A i, dpa iq # the derivatives acting on the external lines corresponding to A i, dimpa iq is the canonical dimension of A i, dimpϕq dimpa µq 1, dimpψq dimp ψq 3 2. If ω ă 0 then in the inductive construction of the time-ordered products the distribution pω T `L ps 1q l 1 px 1q,..., L ps n`1q l n`1 px n`1q Ωq (13) is determined uniquely by the time-ordered products with at most n arguments. 9 / 16
Existence of the weak adiabatic limit Normalization condition which implies the existence of the weak adiabatic limit The weak adiabatic limit exists if for all super-quadri-indices s 1,..., s k which involve only massless fields the time-ordered products satisfy the condition pω Tp L ps 1q l 1 pq 1q,..., L ps kq l k pq k qqωq p2πq 4 δpq 1 `... ` q k q Σpq 1,..., q k 1 q, (14) where B γ q Σp0q 0 for all multi-indices γ such that γ ă ω, i.e. Σ has zero of order ω at zero. The value of the distribution B γ q Σ at zero is defined in the sense of Łojasiewicz. In the case of the above-mentioned class of models it is always possible to define the time-ordered products such that they satisfy the above condition. Comments: 1. According to the above condition the photon self-energy corrections have zero of order 2 in the sense of Łojasiewicz at vanishing external momentum. 2. The correct mass normalization of all massless fields (= vanishing of the self-energy at vanishing external momentum) is necessary for the existence of the weak adiabatic limit. 3. Since the correct mass normalization is not possible in the massless ϕ 3 theory, the weak adiabatic limit does not exist in this theory (the massless ϕ 3 theory does not satisfy the assumption of the theorem from the previous slide). 10 / 16
Compatibility of normalization conditions The normalization condition given on the previous slide is compatible with all the standard normalization conditions: 1. unitarity, 2. Poincaré covariance, 3. CPT covariance, 4. field equations, 5. Ward identities in the QED. Almost homogenous scaling in purely massless models Let A r 1,..., A r k be monomials built out of massless fields. Then pω TpA r 1 px 1q,..., A r k px k qqωq (15) scales almost homogeneously with degree kÿ D dimpa r j q. (16) j 1 11 / 16
Outline of the proof The Wightman and Green functions are formal power series in the coupling constant e ż d 4 x 1... d 4 x m GpB 1px 1q,..., B mpx mqq fpx 1,..., x mq 8ÿ k 0 e k ż d 4 x 1... d 4 x m G k pb 1px 1q,..., B mpx mqq fpx 1,..., x mq. (17) The coefficients of the formal power series are obtained by taking the adiabatic limit ż lim d 4 y 1... d 4 y k d 4 x 1... d 4 x m g ɛpy 1q... g ɛpy k q fpx 1,..., x mq ɛœ0 pω RpLpy 1q,..., Lpy k q; B 1px 1q,..., B mpx mqqωq. (18) The above limit (if exists) is equal to the value at zero in the sense of Łojasiewicz of the following distribution rpq 1,..., q k q : ż d 4 p 1 p2πq 4... d4 p m p2πq 4 fpp 1,..., p mq pω Rp Lpq 1q,..., Lpq k q; B 1p p 1q,..., B mp p mqqωq. (19) 12 / 16
Regularity of a distribution near the origin Notation tpq, q 1 q O dist p q δ q generalizing notation due to Estrada (1998) Let t P S 1 pr N ˆ R M q. For δ P R we write tpq, q 1 q O dist p q δ q, (20) iff there exist a neighborhood O of the origin in R N ˆ R M and a family of functions t α P CpOq indexed by multi-indices α such that 1. t α 0 for all but finite number of multi-indices α, 2. t αpq, q 1 q ď const q δ` α for pq, q 1 q P O, 3. tpq, q 1 q ř α Bα q t αpq, q 1 q for pq, q 1 q P O. Properties 1. If t P CpR N ˆ R M q such that tpq, q 1 q Op q δ q, i.e. tpq, q 1 q ď const q δ in some neighborhood of the origin in R N ˆ R M, then tpq, q 1 q O dist p q δ q. 2. If t P S 1 pr N ˆ R M q and tpq, q 1 q O dist p q δ q then tpq, 0q O dist p q δ q. 3. If t P S 1 pr N q and tpqq c ` O dist p q δ q for some c P C and δ ą 0 then t has value c at zero in the sense of Łojasiewicz. 13 / 16
Outline of the proof (continued) Theorem rpq 1,..., q k q ż d 4 p 1 p2πq 4... d4 p m p2πq 4 fpp 1,..., p mq has value at q 1... q m 0 in the sense of Łojasiewicz. pω Rp Lpq 1q,..., Lpq k q; B 1p p 1q,..., B mp p mqqωq. (21) s ps 1,..., s k q, r pr 1,..., r mq lists of super-quadri-indices involving only massless fields ż r s,r pq 1,..., q k ; q1, 1..., qmq 1 d 4 p 1 : p2πq... d4 p m 4 p2πq fpp 1,..., p mq 4 pω Rp L ps1q pq 1q,..., L pskq pq k q; B pr 1q 1 pq1 1 prmq p 1q,..., B m pqm 1 p mqqωq (22) Induction hypothesis There exists c P CpR 4m q such that for all ɛ ą 0 it holds r s,r pq 1,..., q k ; q 1 1,..., q 1 mq cpq 1 1,..., q 1 mq ` O dist p q 1,..., q k ω1 ε q, (23) where pÿ ω 1 1 rdimpa iq epa iq ` dpa iqs. (24) i 1 The distribution d s,r pq 1,..., q k ; q 1 1,..., q 1 mq is defined in analogous way to the distribution r s,r pq 1,..., q k ; q 1 1,..., q 1 mq with the product R replaced by D A R. 14 / 16
Outline of the proof (continued) It is possible to represent d s,r pq 1,..., q k ; q 1 1,..., q 1 mq with k n in terms of Wightman two point functions of free fields, r s,r pq 1,..., q k ; q 1 1,..., q 1 mq with k ă n and pω Tp L ps 1q pq 1q,..., L ps kq pq k qωq p2πq 4 δpq 1 `... ` q k q Σ s pq 1,..., q k 1 q. s ps 1,..., s k q, r pr 1,..., r mq lists of super-quadri-indices involving only massless fields The proof of the inductive step is divided into two parts: 1. Using the above representation and the lemma below we first show that for all ɛ ą 0 d s,r pq 1,..., q k ; q 1 1,..., q 1 mq O dist p q 1,..., q k ω1 ε q. (25) Lemma It is possible to normalize the time-ordered products such that for all super-quadri-indices s 1,..., s k which involve only massless fields and all ɛ ą 0 where Σ s pq 1,..., q k 1 q O dist p q 1,..., q k 1 ω ε q, (26) pÿ ω 4 rdimpa iq epa iq ` dpa iqs. (27) i 1 2. Next, using the above result we prove that for all ɛ ą 0 r s,r pq 1,..., q k ; q 1 1,..., q 1 mq cpq 1 1,..., q 1 mq ` O dist p q 1,..., q k ω1 ε q. (28) 15 / 16
Other results and open problems Existence of the central splitting solution in the QED For all polynomials B 1,..., B k which are sub-polynomials of the interaction vertex the retarded product satisfy the condition pω Rp B 1pq 1q,..., B npq nq; B n`1pq n`1qqωq p2πq 4 δpq 1 `...`q n`1q rpq 1,..., q nq, (29) where B γ q rp0q 0 for all multi-indices γ, such that γ ď ω ř n`1 j 1 dimpbjq. ñ It fixes uniquely the time-ordered products of sub-polynomials of the interaction vertex. ñ It implies the standard normalization conditions e.g. the Ward identities. The existence of the Wightman functions may be used to define a Poincaré invariant functional on the algebra of interacting fields. This functional is a positive (ñ it is a state) in the case of models without vector fields. Is it possible to define the Poincaré invariant state on the algebra of observables in the QED or the non-abelian Yang-Mills theories without matter? The financial support of the Polish Ministry of Science and Higher Education, under the grant 7150/E-338/M/2017, is gratefully acknowledged. 16 / 16